Objective: To compare the Neuropsychiatric Inventory apathy subscale (NPIa) with the abbreviated Apathy Evaluation Scale (AES-10) on discriminant validity and on their performance to distinguish residents as apathetic or nonapathetic.Design: Cross-sectional design.Setting: Nursing home.Participants: 100 residents of 4 dementia special care units (n = 58) and 3 somatic units (n = 42).Measurements: Primary professional caregivers were interviewed to score the AES-10 and NPIa. The elderly care physician and the psychologist of each unit examined residents for clinical apathy using diagnostic criteria.Results: The AES-10 and NPIa correlated moderately with each other (r(s) = 0.62, P < .0001). The AES-10 correlated weakly (r(s) = 0.27, P = .024) and the NPIa moderately (r(s) = 0.46, P = .001) with the Cornell Scale for Depression in Dementia. Receiver operating characteristic analysis showed an area under the curve (AUC) of 0.72 (P < .01) for AES-10 and 0.67 (P < .05) for NPIa. The AES-10 produced higher sums of sensitivity and negative predictive value than theNPIa. Explorative analyses revealed that both instruments produced higher scores in dementia independently of having an apathy diagnosis, whereas AUCs were significant in nondementia (AES-10: AUC - 0.88, P < .001; NPIa: AUC - 0.77, P = .023), but not in dementia.Conclusion: Both the AES-10 and NPIa may be used to distinguish apathetic from nonapathetic residents in a heterogeneous sample with and without dementia, or in residents without dementia. The AES-10 may be preferable to the NPIa apathy subscale when ruling out or screening for apathy. The performance of the scales against diagnostic criteria of apathy in dementia need to be further examined. Copyright (C) 2012 - American Medical Directors Association, Inc.